🔹Project name: RAG-Based Systems for Information Retrieval and Technical Drawings Analyses
🔹Project’s period: January 2025 – December 2025
🔹Partner: ŠVEC Group
The collaboration centred on identifying and implementing suitable AI solutions to improve ŠVEC Group’s internal processes. The scope evolved over the course of the project as individual use cases were assessed for feasibility and practical value.
HOW WE APPROACHED IT
The project progressed through several directions, each evaluated on its own merits:
🔹Internal data chatbot: One of the initially considered options was the creation of a chatbot to work with company data and support reporting activities. Following an initial analysis, this use case was not implemented, as the company’s data structure was not sufficiently prepared and the expected outputs would not have been fully reliable.
🔹SharePoint document chatbot: The focus then shifted to developing a chatbot for working with documents stored in Microsoft SharePoint. We implemented and tested a solution concept using both custom Python-based tools and the Microsoft Copilot interface. The partner did not see a clear practical benefit for everyday use and decided to discontinue this direction.
🔹Technical drawing analysis: With the remaining project capacity, the use of AI for the analysis of technical drawings and the subsequent generation of import files was explored. Based on manuals and input materials provided by ŠVEC Group, we developed a solution concept whose results met the partner’s expectations.
“The project demonstrated the importance of iteratively validating AI use cases in real conditions. While some directions proved impractical, the technical drawing analysis showed clear potential and measurable impact. This approach helped ensure that the final solution was both feasible and aligned with the partner’s operational needs.“
Ing. TIMOTEJ KRÁLIK
Research Engineer in NLP team
WHAT WE DELIVERED
Our final output was a solution concept delivered in a cloud-based environment using Deepnote. The solution records generated results and stores them in the partner’s SharePoint system, allowing them to be further used in the company’s internal processes.
“The project helped us to separate a few attractive AI ideas from solutions with real operational value. Through several tested directions, we gained a clearer understanding of our data readiness, our internal needs, and the areas where AI can bring practical benefits to ŠVEC GROUP. I really liked the iterative and friendly KINIT approach and how we opened space for further collaboration. The project resulted in a 15% time saving for a specific part of the technical production preparation process in one of the divisions, allowing employees to focus on more creative tasks.”
Mgr. PETER BALLON, MSc., PhD.
PMO Manager, ŠVEC GROUP
